The article analyzes to which extent residential proximity from an emergency hospital affects the probability of surviving an acute myocardial infarction (AMI). The critical time aspect in AMI treatment provides an ideal application for evaluating this proximity-outcome hypothesis. Previous studies have encountered empirical difficulties relating to potential endogenous health-based spatial sorting of involved agents and data limitations on out-of-hospital mortality. Using policy-induced variation in hospital distance, arising from plausibly exogenous emergency hospital closures in the highly regulated Swedish health care sector, and data on all AMI deaths in Sweden over two decades, estimation results show a clear robust and gradually declining probability of surviving an AMI of about two percentage points (three percent) per additional ten kilometers distance from a hospital. The results further show that spatial sorting and sample selection from out-of-hospital mortality are likely to significantly attenuate the distance effect unless accounted for.